ONC’s first draft of a nationwide interoperability roadmap is ambitiously vast in scope, but ultimately constrained by the past. Its purpose is to initiate a discussion within healthcare about the ways and means to achieve interoperability in 10 years notwithstanding that the discussion is already the obsession of many. It hopes that this document will launch a process that results in a national private-public strategy for supporting the kind of interoperable data infrastructure that will enable a learning health system.
The roadmap focuses on several major policy and technical themes: the impact of FFS on vendor and provider attitudes to data sharing, the potential for private payers and purchasers to incentivize data sharing, the central importance of standards and incentivizing compliance. This long, comprehensive, and in places incredibly detailed, document is really three documents in one. For those lacking the time to read through its 160+ pages, we summarize:
Part I – Letter from ONC chief Karen DeSalvo and executive summary that lays out a set of questions intended to guide the response to the overall document as well as a series of “calls to action” to galvanize industry participation. (Pages 1-15)
Part II – Exhaustive presentation of the current and potential future state of interoperability, as well as the challenges and opportunities that lay between. (Pages 6-162)
Part III – The final appendix containing 56 “Priority Interoperability Use Cases” which ONC wants to winnow down. (Pages 163-166)
Part I gives a sense of which way ONC is leaning by focusing on select high-level issues: payment policy, data governance, semantic and transactional standards, measurement of results, and probably most important, the priority use cases. The language used within the roadmap is reflective of a sea change in thinking: “send and receive” has been replaced with “send, receive, find, and use” as a way to describe what individuals need from interoperable HIT.
Part II lays out the elements of what it hopes will become the national roadmap for interoperable HIT. The roadmap’s focus on measurement – tracking and gauging the metrics of interoperability – is eerily similar to the EHR Incentive Program’s MU measures right down to using various numerators and denominators. If we learned anything from the EHR Incentive Program it is that the industry liked the incentives but disliked the actual MU objectives. Absent the carrot of incentives and/or the sting of penalties, it is hard to see how ONC can catalyze providers to embrace yet another set of operational metrics.
ONC continues to struggle with patient matching. We know that Congress will not countenance so much as a voluntary (on the part of patients) national patient identifier. Even if it did, the costs to the industry of maintaining a dual system would only add complexity to an overburdened system. Unfortunately, we think that ONC and HHS is powerless to change this extravagantly costly element of the interoperability conundrum.
Who Uses What Data?
Embedded in ONC’s treatment of HIPAA, data governance, and data portability lies an essential, unresolved issue: The rights and responsibilities of the various stakeholders with respect to the data governance. How exactly can personal health information (PHI) that is captured by clinical applications be shared within the context of care delivery?
Today, such rules of the road remain unclear, ambiguous, and deeply complex. Existing laws, both state and federal, are a patchwork in which various participants hold back data fearing liability where most often no liability exists. At the same time, various participants claim outright to ”own” this data and use this patchwork to consolidate their competitive position.
ONC rightly points out that ATM networks and airline reservation networks provide interoperability for radically simpler use cases than the health system requires (they also don’t quite have the regulatory complexity of health data). While true, the data practices of consumer-focused transaction networks are effectively incomprehensible for the average consumer. So why should we assume that therefore somehow how we need to make it comprehensible for this use case of data?
Is it reasonable to expect any patient to understand the protections offered in the vastly more complex health data realm? Resoundingly no. ONC could be soliciting input about comprehensive reconceptualization of data rights and responsibilities with respect to patient data. Admittedly, only Congress can act to change the existing regime, and even Congress is limited by legislation enacted at the state level.
This is one aspect of interoperability that will aways confound those seeking true interoperability and data exchange in the context of health. It is also why some proponents believe that interoperability will only truly occur once the patient has full control of their data (health data bank) and defines access to their PHI. But even here, we have a very long way to go before the majority of citizens take upon that responsibility.
Standards and Compliance
Most of the ideas embodied in the JASON Report and the subsequent JASON Task Force are offered up for public comment. The roadmap suggests that HL7’s new standard, FHIR, could be effective in the 6-10 year timeframe, considerably longer than the time contemplated by the Argonaut Project. ONC also stops short of saying that element-centric interoperability will or even should replace document-centric exchange. But it talks about electronic sharing of summary care records – not documents – between hospitals, SNFs, and home health agencies.
The roadmap reinforces and effectively doubles down on the centrality of standards in any plan to foster better interoperability. Borrowing liberally if not literally from the Common MU Data Set, ONC wants to know how it can help make data more intelligible inside and between providers. As a companion to the roadmap it also released an “Advisory” on the most common standards in order to get opinions on where the industry’s best practices focus should be. ONC believes that standards-compliance and the elimination of non-conforming data representations will pay dividends.
The counterpoint to this emphasis on standards is the view held by many smaller, often start-up vendors that see standards as a means to preserve the status quo, serving more of gatekeeping function than an enabling function. Data networks in other industries, while admittedly simpler, do not rely on prescriptive application-to-application data representation standards. Healthcare is the only industry with such an ornate implementation of level 7 of the OSI stack. Smaller vendors would rather see simpler standards, published APIs, or more of a focus on the results of exchange than on how the result is to be achieved.
The reality is that major HIT vendors and major providers grumble about prescriptive requirements but by and large remain deeply committed to standards and compliance. We think that ONC could have at least offered up the prospect of achieving interoperability goals without specifying the mechanism down to specific data elements. Unfortunately, ONC appears to be continuing upon its questionable path of highly prescriptive guidelines – guidelines that ultimately hinder innovation rather than create opportunities for innovation to flourish.
Read the Priority Use Cases Right Now
Part III contains 56 uses cases, obviously culled from users, and are a dog’s breakfast of highly specific and extremely general interoperability complaints. ONC is asking the industry to help it order and prioritize the vast range of technical and policy question that they raise.
We recommend that if you read nothing else in the entire roadmap, you should read these use cases because they sound more like demands than actual use cases.
For example, Item #8 – certified EHR technology (CEHRT) should be required to provide standardized data export and import capabilities to enable providers to change software vendors. Every HIT vendor believes publishing database schemas is the last stop on the way to mob rule. In case vendors as a class were uncertain about their reputation among at least some providers, this “use case” provides unambiguous feedback.
A large number of the use cases are decidedly patient-centric despite the decidedly provider-centric orientation of the wider healthcare system and significant resistance to any kind of reorientation. A significant number of the use cases are related to payment and administrative uses even though the roadmap’s focus is on clinical data and clinical uses of the data. There are also a large number of use cases related to support for the public health system and clinical research. Both of these constituencies unfortunately take a back seat to immediate patient care and payment priorities.
The roadmap mentions, without analysis, the fundamental problem of FFS-based disincentives to data sharing. HHS has recently announced new goals for progress on the way to VBR but ONC has little leverage to do much more.
Another important issue that the roadmap does not and likely can’t address is the level of investment in IT by healthcare providers. While many yearn for an interoperable infrastructure comparable to what banking or retail enjoy, those industries spend far more, as a percentage of revenue, on IT than healthcare providers. Progress on EHR adoption was not a result of provider’s reallocating resources to technology adoption, but federal incentives under the HITECH Act.
Therefore, can we really expect HCOs to increase IT budgets to support interoperability? Probably not. Moreover, ONC and more broadly HHS, do not have the funding to support interoperability adoption on the scale of EHR adoption via the HITECH Act, absent congressional action. Most HCOs are cash-strapped and struggling with a multitude of changes occurring in the marketplace and frankly have a fairly poor record in the effective adoption, deployment and use of IT in the context of care delivery. This is a knowledge intensive industry that has done a pretty lousy job at effectively harnessing that knowledge via IT.
The only leverage ONC and HHS have to improve interoperability is payment incentives via CMS. Recently, HHS announced that it will accelerate the move to VBR. Following closely on that announcement was the formation of the Healthcare Transformation Task Force, an industry association that sees its task as helping the industry migrate to VBR. It is far more likely that organizations such as this in conjunction with payment reform will do far more to achieve interoperability than any prescriptive roadmap.
It may be high time for ONC to step back and let the industry tackle this one on their own for only they will truly have a vested interest, via payment reform, to share PHI in the context of care delivery across a community in support of the triple aim and population health management.
I was struck by the criticisms of ONC for overly prescriptive guidelines:
“Unfortunately, ONC appears to be continuing upon its questionable path of highly prescriptive guidelines – guidelines that ultimately hinder innovation rather than create opportunities for innovation to flourish.”
I understand why this is viewed as problematic by the industry for facilitating innovation, but I think the attitudes of many people, and in particular many clinicians and patients, drove ONC in this direction. In healthcare IT the expectations for grappling with complex workflows and subject matter and reams of detail are very high, and systems are likely to be slammed for being simplistic and insensitive to the needs and expectations of this or that constituency. As a result, any agency that presumes to impose some order on the confusion will understandably attempt to forestall various classes of objections by prescribing specific solutions and methodologies.
If this ends up coming to grief it is very often because the various constituencies are in conflict with each other. They don’t want to deal with other, and probably more often that not they don’t think they can. In either case they want someone in authority to advocate for their needs and concerns.
So, this evolutionary process is inevitably a painful and protracted one. The various constituencies are slowly figuring out what compromises (real or apparent) to accept, and the benefits of coming to agreement around the various issues in play. One can’t see why this has to happen unless one sets aside one’s own concerns for a while and simply surveys the pressures on the numerous participants and stakeholders — usually generated by other participants and stakeholders.
ONC gave the constituencies what they asked for and is now dealing with the aftermath. The expectation level for many front-line clinicians is, for better or worse, driven or at least influenced by their personal experiences with consumer websites. It is easy to point to some credit card website and say, “see…they know every place I bought something last month,” and wonder why something like it can’t happen at a nurse station say. Balancing the demands of the various groups in light of their need and understanding of what is possible is a dicey proposition and I don’t envy ONC or anyone responsible for any healthcare standard. The link in the Walter Sujansky’s comment fully captures this issue.
Brian makes some excellent points about the strengths and limitations of the interoperability roadmap. Notably, that providers themselves must come to demand and pay for effective interoperability features as a condition of their purchase decisions. This is critical so that EHR vendors will collectively perceive the benefits of providing interoperability as higher than the costs. At the same time, the costs must also be reduced for both vendors and providers through the creation of more specific, more constrained, and better tested standards. The historical inability of our industry to create such standards results from a fundamentally flawed approach (anyone interested can see a more detailed discussion of this at http://www.ihealthbeat.org/perspectives/2015/interoperability-failure-to-launch). Until both the benefit and cost sides of the equation are addressed, I fear that interoperability will remain 10 years away for many more decades…
Making HIT more interoperable is being approached more like a renovation project than new construction. The drawbacks of this approach that you highlight in your excellent and highly recommended iHealthBeat piece are on point. Healthcare unquestionably suffers from too many cooks working earnestly but at cross-purposes. As you also suggest, the whole process for moving this stuff into production is rushed or haphazard or both. Too often the people involved have clear ideas of what the benefits will be (although every one has different ideas of the specific set of benefits) and have hazy ideas of what the immediate and long-term costs will be. I may be Pollyannaish but better interoperability in healthcare is possible and I don’t think we’ll have to wait that long.
[…] recent Chilmark blog post on national healthcare interoperability mentioned two approaches to healthcare interoperability: […]
Brian , I am trying to understand the comment: “ONC also stops short of saying that element-centric interoperability will or even should replace document-centric exchange.”
Page 82 of the “A Shared Nationwide Interoperability Roadmap” states that data element exchange will overtake document centric methods.
Here is the excerpt “Over the long-term, document-centric ways of exchanging electronic health information will likely be overtaken by more specific, tailored, data-centric ways of exchanging information – meaning that individual data elements are exchanged ( e.g., medications or allergies) rather than an entire document with multiple information sections about a patient.”
Doesn’t that mean ONC expects element centric interoperability will be the final state? Let me know your thoughts.
Can I suggest that the key word here is “overtake” and not “replace”? There are still an extreme numbers of faxes and phone calls circulating within and between HCOs, most of which could be replaced with document- or element-centric exchange. To your point, element-centric interoperability is a far better solution along multiple dimensions. Document-centric exchange is an embedded practice, it works in many places, and will, sadly, persist for a long time.